The R package 'adagio' will provide methods and algorithms for discrete optimization, e.g. knapsack and subset sum procedures, derivative-free Nelder-Mead and Hooke-Jeeves minimization, and some (evolutionary) global optimization functions.
Changes in Version 0.7.1 (2018-05-16)
o Removed 'adagio-package.Rd' on request of K. Hornick, CRAN.
Changes in Version 0.7.0 (2017-06-28)
o pureCMAES() added a break statement, proposed by R. Biedrzycki.
Changes in Version 0.6.9 (2017-02-07)
o nelmin(), nelminb() renamed to neldermead(), neldermeadb(). o hookejeeves() Hooke-Jeeves minimization (with bound constraints). o Help page of knapsack() did not mention the interger requirement.
Changes in Version 0.6.7 (2017-02-06)
o nelmin(), nelminb() now use an dimension-adaptive Nelder-Mead minimization by F. Gao and L. Han. o simpleDE() returns the number of function evaluations.
Changes in Version 0.6.5 (2016-05-29)
o corrected an error in grNesterov(), as indicated by John Nash.
Changes in Version 0.6.3 (2015-12-06)
o hamiltonian() finds a Hamiltonian path or cycle in a graph given as a vector of edges (see the 'igraph' package) or an edge list.
Changes in Version 0.6.1 (2015-07-25)
o Reorganized test functions; removed genTestfn() and (re-)introduced Rosenbrock, Nesterov, Rastrigin, Hald, Shor with their true names; prepended with a 'fn' for the function, 'gr' for existing gradients. o Removed linesearch_ww(), linesearch_sw(), and qpspecial() for now. o trefethen(), wagon() renamed to fnTrefethen(), fnWagon(). o Corrected a call to maxempty() on the help page.
Changes in Version 0.5.9 (2013-10-27) o fminviz(), flineviz() visualize multivariate functions at (assumed) minima or along lines in multidimensional space.
Changes in Version 0.5.7 (2013-09-12) o pureCMAES() Covariance Matrix Adaptation Evolution Strategy with applications to the Remez problem and L1 regression.
Changes in Version 0.5.4 (2013-08-09) o genTestfn() generates optimization test functions: Hald, Nesterov, Nesterov1, Rastrigin, Rosenbrock, Shor. o maxquad() Lemarechal's MAXQUAD test function generator.
Changes in Version 0.5.1 (2013-06-30) o linesearch_ww(), linesearch_sw() Wolfe's weak and strong line search. o qpspecial() special quadratic programming solver.
Changes in Version 0.4.3 (2012-11-11) o nelmin(), nelminb() Nelder-Mead function minimization according to O'Neill (Fortran code) and J. Burghardt (Matlab code, LGPL licence). o transfinite() transformation function between bounded and unbounded (box constrained) regions.
Changes in Version 0.3.9 (2012-06-19) o simpleEA() and simpleDE() improved, with boundary conditions. o simpleEA() speed improved, with new parameters and options.
Changes in Version 0.3.7 (2012-03-24) o simpleDE() simple differential evolution algorithm for function minimization.
Changes in Version 0.3.5 (2012-03-17) o assignment() solves the linear (sum) assignment problem (LSAP) for integer square cost matrices.
Changes in Version 0.3.3 (2012-03-14) o maxempty() solves the maximal empty rectangle problem.
Changes in Version 0.3.1 (2012-03-12) o mknapsack() solves the multiple knapsack problem, utilizing free Fortran code from the NSCW library of Mathematics Subroutines.
Changes in Version 0.2.8 (2012-02-16) o knapsack() solves the 0-1 (binary) single knapsack problem using the dynamic programming approach. o subsetsum() now with 'greedy' and 'dynamic' methods.
Changes in Version 0.2.5 (2012-01-25) o maxsub2d() maximal sum subrectangle problem; uses Fortran code to be able to handle 1000-by-1000 matrices.
Changes in Version 0.2.0 (2012-01-14) o maxsub() maximal sum subarray (with Ratfor and Fortran code). o subsetsum() Subset sum routine written in R.
Changes in Version 0.1.6 (2012-01-07) o Set up as package on R-Forge, part of the Optimist project.
Changes in Version 0.1.5 (2011-07-27) o simpleEA(), simple evolutionary algorithm for minimum finding. o Test functions trefethen() and wagon(). o Setting up the package structure and required files.